Intra-individual psychological and physiological responses to acute laboratory stressors of different intensity

Psychoneuroendocrinology (2015) 51, 227—236

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/psyneuen

Intra-individual psychological and physiological responses to acute laboratory stressors of different intensity Nadine Skoluda a, Jana Strahler a, Wolff Schlotz b, Larissa Niederberger c, Sofia Marques c, Susanne Fischer a, Myriam V. Thoma c, Corinne Spoerri c, Ulrike Ehlert c, Urs M. Nater a,∗ a

University of Marburg, Department of Psychology, Germany Max Planck Institute for Empirical Aesthetics, Germany c University of Zurich, Department of Psychology, Switzerland b

Received 6 June 2014; received in revised form 10 September 2014; accepted 1 October 2014

KEYWORDS Cortisol; Alpha-amylase; Heart rate; Stress reactivity; Laboratory stressor; Trier Social Stress Test; Ergometer; Cold pressor test; Stroop test

Summary Objectives: The phenomenon of stress is understood as a multidimensional concept which can be captured by psychological and physiological measures. There are various laboratory stress protocols which enable stress to be investigated under controlled conditions. However, little is known about whether these protocols differ with regard to the induced psycho-physiological stress response pattern. Methods: In a within-subjects design, 20 healthy young men underwent four of the most common stress protocols (Stroop test [Stroop], cold pressor test [CPT], Trier Social Stress Test [TSST], and bicycle ergometer test [Ergometer]) and a no-stress control condition (rest) in a randomized order. For the multidimensional assessment of the stress response, perceived stress, endocrine and autonomic biomarkers (salivary cortisol, salivary alpha-amylase, and heart rate) were obtained during the experiments. Results: All stress protocols evoked increases in perceived stress levels, with the highest levels in the TSST, followed by Ergometer, Stroop, and CPT. The highest HPA axis response was found in the TSST, followed by Ergometer, CPT, and Stroop, whilst the highest autonomic response was found in the Ergometer, followed by TSST, Stroop, and CPT.

∗ Correspondence to: University of Marburg, Clinical Biopsychology, Gutenbergstrasse 18, 35032 Marburg, Germany. Tel.:+49 6421 28234943; fax: +49 6421 28 24077. E-mail address: [email protected] (U.M. Nater).

http://dx.doi.org/10.1016/j.psyneuen.2014.10.002 0306-4530/© 2014 Elsevier Ltd. All rights reserved.

228

N. Skoluda et al. Conclusions: These findings suggest that different stress protocols differentially stimulate various aspects of the stress response. Physically demanding stress protocols such as the Ergometer test appear to be particularly suitable for evoking autonomic stress responses, whereas uncontrollable and social-evaluative threatening stressors (such as the TSST) are most likely to elicit HPA axis stress responses. The results of this study may help researchers in deciding which stress protocol to use, depending on the individual research question. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction When studying the phenomenon of stress under controlled conditions, researchers need standardized laboratory stress protocols which induce stress in a reliable and valid manner. Such protocols will enable the researcher to investigate (a) the psychological and physiological mechanisms of the stress process itself, and (b) the emotional, cognitive, and behavioral consequences of induced stress. There are various approaches to describe the domains of stress responses, the physiological domain, the subjective experiences domain (including emotional states), the cognitive domain, and the behavioral domain. In the following, we focus on a bi-modal approach, according to which the stress response comprises both psychological and physiological processes. The ‘Transactional Model of Stress’ (Lazarus and Folkman, 1984) focuses on psychological mechanisms and proposes that an imbalance of primary appraisal (i.e., evaluation of the current situation as threatening and potentially harmful) and secondary appraisal (i.e., evaluation of an individual’s resources to cope with threat) results in a stress response. The central nervous system (CNS) plays a key role in these psychological processes and also orchestrates the body’s response to a stressor. Stress-induced CNS activation drives the two most prominent physiological stress-response systems, namely the hypothalamus—pituitary—adrenal (HPA) axis and the autonomic nervous system (ANS), particularly the sympatho—adrenal—medullary (SAM) system (Chrousos, 2009). Due to its quick and uncomplicated assessment, the most commonly used marker of HPA axis responses is salivary cortisol. Autonomic stress responses are frequently obtained by non-invasive measurement such as heart rate (HR) or salivary alpha-amylase (sAA) activity. Altered psycho-physiological stress responses to different laboratory stressors are related to various psychiatric and somatic disorders (Gerra et al., 2001; Hamer and Steptoe, 2012; Kelly and Cooper, 1998). Laboratory stressors may thus be used as a tool for predicting and/or diagnosing negative health outcomes. Variability in stress responses can be attributed to the factors originating from the person or the environment. The concept of stimulus response specificity reflects the observation that variability in stress response patterns is to some degree attributable to situational characteristics or the type of stressor a person is exposed to (Schlotz, 2013). For example, it has been suggested that HPA axis responses are specifically triggered by social evaluative threat (Dickerson and Kemeny, 2004), whereas stressors characterized by effort might specifically trigger SAM responses (Lundberg and Frankenhaeuser,

1980). Alternatively, it has been suggested that variability in response pattern among stressors might essentially be attributable to differences in response intensity elicited by different stressors (Bosch et al., 2009). Regardless of the specific mechanisms underlying response differences associated with stressor types, it is obvious that detailed information about such patterns is useful for decisions on which stressor to employ in a stress study. The primary aim of this study therefore was to compare psycho-physiological stress response patterns of various stressors and a non-stress resting condition in a within-subjects design. The specific stressors studied here were chosen on the basis of their popularity in stress research as well as their potency to elicit stress responses (Biondi and Picardi, 1999). We searched the literature and reviewed previous peerreviewed studies which (a) used at least two different laboratory stress protocols, and (b) reported a measure of psychological stress response (perceived stress) or reported findings on measures of the physiological stress response (cortisol, sAA, or HR), or both (see supplementary material I). The review revealed that the majority of stressors indeed successfully induced psycho-physiological stress responses indicated by various stress measures. However, the pattern of stress responses associated with a specific stressor was rather inconsistent across studies. This inconsistency might be explained by methodological limitations within and differences between the reviewed studies. In most of the studies, participants were exposed to multiple stressors during the same session, which does not allow for a direct comparison between single stressors due to potential carry-over effects. This selection resulted in only a small number of remaining studies. A list of these studies with details on methods, design, and outcome measures is provided as a supplement (see supplementary material I). It becomes evident from this list that the available studies allow only limited conclusions about the comparability of stressors, as most suffered from various methodological limitations, such as between-subjects designs (different participants were exposed to different stressors), comparison of no more than two stressors in a within-subjects design, or a lack of control condition (no comparison between stress and no-stress condition). Furthermore, the majority of studies focused on a single stress measure, thus ignoring the multidimensional nature of stress. In summary, there is a lack of studies using a within-subjects design which systemically compare psycho-physiological stress responses between various stress protocols, and provide a comparison with a no-stress control condition. The current study was designed to overcome these methodological limitations and one-dimensional assessment of stress responses.

Stressor intensity & psycho-physiological responses The aim of the current study was to compare psychological and physiological stress responses between four commonly used stress protocols using a within-subjects design: Stroop test (Stroop), Trier Social Stress Test (TSST), cold pressor test (CPT), and bicycle ergometer test (Ergometer), as well as a no-stress control condition (rest). We hypothesized that these stressors would differ in their magnitude and pattern of psycho-physiological stress response, namely (a) self-reported perceived stress, (b) salivary cortisol concentration, (c) sAA activity, and (d) HR. In addition, we expected to find differences in average primary and secondary appraisal elicited by the specific stressors. Due to inconsistencies in previous studies, it was not possible to provide specific ranking hypotheses. A further aim of our study was to establish a ranking order of stress protocols indicating the relative magnitude of stress responses in different domains.

229 first saliva sample and completed a visual analogue scale assessing momentary perceived stress levels (VAS stress). Subsequently, participants were taken to room 2, where they were briefly introduced to the respective stress task. Following this introduction, participants returned to room 1, where they completed questionnaires (PASA and VAS stress) for 10 min (anticipation or pre-stress period). Immediately before performing the respective stress task in room 2 (or remaining in room 1 in the no-stress condition), a second saliva sample was collected. After completing the stress task, participants returned to room 1 and a third saliva sample and momentary perceived stress levels (VAS stress) were obtained (post-stress period). In the remaining time, five additional saliva samples were collected (recovery period). At the end of each session, participants were debriefed, thanked, and dismissed.

2.3. Conditions

2. Methods In the following, the five conditions are described in detail:

2.1. Participants Twenty healthy young men (mean age: 23.1 ± 2.4 SD, range: 19.5—27.4 years; BMI: 23.2 ± 2.7 SD, range: 18.5—28.7 kg/m2 ) were recruited via various advertising outlets among students enrolled at the University of Zurich. In a telephone-based interview potential subjects were screened regarding inclusion and exclusion criteria. Exclusion criteria were female gender (as the female menstrual cycle is known to affect neuroendocrine stress responses), smoking, drug use, use of medication known to affect endocrine and autonomic functioning, regular alcohol use (>3 alcoholic drinks per day), psychiatric and somatic diseases known to affect endocrine functioning, and color blindness. Participants received monetary compensation (100 SFR) or course credit. After receiving written and oral information about the study, written informed consent was provided by all participants. The study protocol was approved by the local ethics committee.

2.2. Study design and procedure In a within-subjects design, each participant underwent five sessions (four stress conditions and one no-stress condition) in a randomized order on five separate days with five to nine days between each session. The session order was randomly selected for each participant separately and before the study entry by using lots with one out of 120 possible session orders. Lots were used only once so that each participant faced an individual session order. Participants were blind to the session order. Sessions took place in the afternoon (between 13.00 and 17.00) to control circadian rhythms of biomarkers assessed in this study. The procedure is illustrated in Figure 1 (for more details, see supplementary material II). Upon arrival, informed consent was obtained and a socio-demographic questionnaire was completed (first session only), and devices for continuous heart rate assessment (Polar S810) were applied. This was followed by a baseline period during which participants rested in room 1 for 30 min in order to adapt to the laboratory setting. At the end of the baseline period, participants provided the

2.3.1. The no-stress control condition (rest) The rest condition served as a control condition in which no stress was induced. Participants were instructed to relax for 10 min. Magazines were provided. 2.3.2. The Stroop color-word interference test (Stroop) The Stroop test (Stroop, 1935) is a cognitive task which requires selective attention and behavioral inhibition of a dominant (re)action. Participants were instructed to name printed colors of items shown on several cards (45 items per card, organized in three rows with 15 items each). In the Stroop version used in this study, there were three types of paper cards: (a) control cards: five circles printed in colors (yellow, red, blue, or green); (b) neutral word cards: neutral words printed in colors; (c) incongruent color-word cards: words of color names were printed in another color. The latter type of cards may result in a conflict between dominant (read the word) and non-dominant reaction (name color of word), which is typically associated with a slower and poorer performance (not measured here). In order to increase the stressfulness of the task, participants were asked to name colors as fast as possible. If participants made a mistake, they were interrupted and asked to name the correct color. 2.3.3. The cold pressor test (CPT) The CPT is a test which is thought to elicit pain and physiological responses associated with both HPA axis and autonomic activity. Participants were instructed to place their non-dominant hand up to the wrist in a box filled with ice-cold water (temperature held constant at 3—4 ◦ C) for as long as possible. They were told that they could remove their hand from the water whenever they wanted. The maximum immersion duration was 3 min (this was not communicated to participants in advance). 2.3.4. The Trier Social Stress Test (TSST) The TSST is a well-established method to induce psychosocial stress in a laboratory setting (Kirschbaum et al., 1993). The TSST in our study consisted of a 5-min mock job interview (instruction to give a free speech in front of an

230

N. Skoluda et al.

Figure 1

Study protocol and procedure. Time indicates the time relative to the condition.

evaluating audience and a video camera) and a 5-min mental arithmetic task (instruction to solve a serial subtraction task). 2.3.5. The bicycle ergometer task (Ergometer) Physical exertion due to exercise is another common laboratory paradigm which reliably induces HPA axis and autonomic stress responses (e.g., Allgrove et al., 2008). In our study, after a 2 min warm-up on a stationary bicycle, participants were asked to continue cycling at a high level (between 80% and 90% of self-rated effort) using the Borg Rating Perceived Exertion (RPE) scale as reference (Borg and Kaijser, 2006) for the remaining 8 min.

2.4. Measures 2.4.1. Primary Appraisal Secondary Appraisal (PASA) The 16-item PASA (Gaab et al., 2005) measures stressrelated anticipatory cognitive appraisal processes based on the Transactional Model of Stress by Lazarus (Lazarus and Folkman, 1984). To determine anticipatory stress appraisal, participants completed the PASA immediately before stress exposure or the rest period at each session. Three indices may be derived: (a) index of primary appraisal (threat and challenge), (b) index of secondary appraisal (self-concept of own abilities, control expectancy), and (c) stress index (difference score between primary appraisal and secondary appraisal). 2.4.2. Visual analogue scale (VAS) stress For the assessment of momentary perceived stress levels, subjects indicated how stressed they felt on a line of 100 mm length. The VAS stress was administered five times during each session: at baseline (−20 min relative to condition), pre-stress (−3 min), post-stress (+1 min), recovery I (+15 min), and recovery II (+30 min). 2.4.3. Salivary cortisol To assess cortisol as a measure of HPA axis activity, saliva samples were collected using Salivettes (Sarstedt, Sevelen, Switzerland). Participants were asked to chew on the swab for 1 min. Saliva samples were stored at −20 ◦ C until biochemical analyses were conducted at the Department of Clinical Psychology and Psychotherapy, University of Zurich. Eight saliva samples were collected during each session, reflecting the free cortisol fraction at baseline (−20 min relative to condition), pre-stress (−3 min), post-stress (+1 min), recovery I (+15 min), recovery II (+30 min), recovery III

(+45 min), recovery IV (+60 min), and recovery V (+75 min). Free cortisol concentration in saliva was determined using a commercially available luminescence immunoassay (LIA, IBL, Hamburg, Germany. Inter- and intra-assay coefficients of variation were below 5%. 2.4.4. Salivary alpha-amylase (sAA) SAA, which is a measure of autonomic activity, was analyzed from the same saliva samples as cortisol. Since sAA is an enzyme, alpha-amylase activity was determined using a kinetic colorimetric test with reagents obtained from Roche (Roche, Basel, Switzerland), with intra- and inter-assay variance below 10%. 2.4.5. Heart rate (HR) Heart rate serves as another measure of autonomic activity. HR (i.e., beats per minute) was continuously recorded during all sessions using Polar Watch devices (S810, Polar Electro GmbH, Büttelborn, Germany). Based on markers set at pre-defined time points, we defined the following time periods: baseline (1—5 min of recording corresponding to 5 min before introduction to task), pre-stress (5—17 min), stress (CPT: 17—20; other conditions: 17—27), and poststress (CPT: 20—32 min; other conditions: 27—39 min). Since the CPT was shorter than all of the other conditions, we selected time periods reflecting pre-stress (5, 8, 11, 15, 16, and 17 min relative to recording onset), stress (CPT: 18, 19, and 20 min; other conditions: 19, 23, and 27 min) and poststress period (CPT: 26 and 32 min; other conditions: 33 and 39 min).

2.5. Data analyses 2.5.1. Missing values One participant provided an insufficient amount of saliva, which resulted in one (rest) and four (CPT) missing values, respectively. We excluded this participant from all statistical analyses. However, including this participant (after imputing missing values by the group mean) did not change any of the findings. Another participant failed to answer one item in the PASA questionnaire in the Stroop condition. This missing value was replaced by the group mean. For HR data, the first 2 min of recording (initially intended to serve as baseline) were missing in 40% of cases due to technical reasons. Thus, we decided not to consider the first 2 min of recording in the statistical analyses. Further missing values were observed in four participants. Specifically, no or incomplete HR data were recorded during the rest (one participant), Stroop (one

Stressor intensity & psycho-physiological responses participant), TSST (three participants), and Ergometer (two participants) condition. These missing values were replaced by the group mean. 2.5.2. Statistical analyses As often observed in biological data, cortisol and sAA data tended to be positively skewed in the current study. Natural logarithm (ln) transformation was conducted, resulting in normally distributed data as statistically confirmed by a Kolmogorov—Smirnov test (all p’s > 0.27). Ln-transformed data were used for statistical analyses, while means and standard errors (SEM) in figures were presented in non-transformed units. Repeated measures ANOVAs (with Greenhouse—Geisser correction in the case of violation of the sphericity assumption) were computed to examine whether conditions (rest, Stroop, CPT, TSST, and Ergometer) differed in baseline values of the respective stress markers (VAS, cortisol, sAA, and HR). Repeated measures ANOVAs were computed to investigate effects of time and condition on stress measures. For comparison between stressors and the creation of a stressor intensity ranking, differences between expected peak and baseline/pre-stress values were calculated (delta values) for each condition. More specifically, delta scores for perceived stress (VAS) and sAA were defined as difference between post and baseline. For cortisol, delta scores were calculated by difference between recovery I and baseline. HR delta scores were determined by the difference of the maximum value during the stress period and the minimum value during the pre-stress period. For comparison regarding these delta values and PASA values, separate repeated measures ANOVAs (with Greenhouse—Geisser correction) and post-hoc tests (pairwise t-tests with Bonferroni correction) were computed to test for overall and specific differences between conditions. For ANOVA analyses, the alpha level was set at < 0.05. Considering cumulative Type I Error due to multiple post-hoc tests (10 pairwise comparisons), we used the Bonferroni correction formula (␣ = alpha level divided by number of tests), resulting in a corrected alpha of < 0.005. As a measure of effect size, we used eta (Á2 ). For the classification of cortisol responders and non-responders, we used the 1.5 nmol/l criterion proposed by Miller and colleagues (2013). Relative responder rates (%) were reported for each stressor separately based on delta scores using expected or individual peaks. All analyses were conducted using SPSS 21.0 (IBM, Chicago, USA).

3. Results 3.1. Psychological and physiological stress measures at baseline To ensure that changes across the conditions were not solely attributable to different baseline levels, baseline values were compared for each stress measure. For the VAS stress, there was a trend for baseline differences (FVAS (2.09, 37.66) = 3.154, p = 0.052, Á2 = 0.149). However, post-hoc tests revealed no significant differences between conditions (all p’s between 0.018 and 0.974). There were no significant differences regarding physiological baseline values

231 (Fcortisol (4, 72) = 0.600, p = 0.664; FsAA (2.58, 46.46) = 1.196, p = 0.319; FHR (4, 72) = 2.188, p = 0.079).

3.2. Time course of stress measures across the conditions Figure 2 depicts the time course of the four stress measures throughout the five conditions. For the VAS, there was a significant effect of time (F(2.88, 51.74) = 15.173, p < 0.001, Á2 = 0.457), indicating changes in perceived stress levels over time across conditions. Furthermore, a significant effect of condition was found (F(1.88, 33.78) = 15.542, p < 0.001, Á2 = 0.463), suggesting that participants’ average stress perception differed by condition. The significant time × condition interaction suggests different time courses of perceived stress among the conditions (F(4.95, 89.14) = 8.956, p < 0.001, Á2 = 0.332). For cortisol, the repeated measures ANOVA revealed a significant effect of time (F(2.32, 41.76) = 17.512, p < 0.001, Á2 = 0.493), a significant effect of condition (F(3.45, 62.03) = 15.599, p < 0.001, Á2 = 0.464), and a significant time × condition interaction (F(6.23, 112.13) = 8.690, p < 0.001, Á2 = 0.326). These findings indicate different HPA axis stress responses among conditions. Similar to cortisol, we found a significant main effect of time for sAA (F(3.92, 70.64) = 19.729, p < 0.001, Á2 = 0.523) but no significant effect of condition (F(3.39, 61.02)=0.819, p = 0.501). However, there was a significant time × condition interaction (F(9.21, 165.70) = 2.856, p = 0.003, Á2 = 0.137), indicating that sAA trajectories differed among conditions. With regard to HR, there was a significant effect of time (F(3.59, 64.66) = 70.657, p < 0.001, Á2 = 0.797), a significant effect of condition (F(2.53, 53.55) = 40.817, p < 0.001, Á2 = 0.694), and a significant time × condition interaction (F(7.94, 142.84) = 61.835, p < 0.001, Á2 = 0.775). We also compared overall differences among conditions across the sessions using the index ‘area under the curve with respect to ground’ (AUCG ) based on the trapezoid formula (Pruessner et al., 2003). It is important to note that AUCG was adjusted by total time duration because the time duration of CPT was shorter than that of the other conditions. Repeated measures ANOVAs using the AUCG confirmed the findings reported above, with significant effects of condition in all stress measures except for sAA (see supplementary material IV).

3.3. Relative stressor intensity To create a stressor intensity ranking, we compared delta values and the three PASA indices between conditions. For the PA index, the repeated measures ANOVA revealed significant differences among conditions (F(4, 72) = 42.169, p < 0.001, Á2 = 0.701). Post-hoc tests showed that all conditions, with an exception of the CPT, were appraised as being significantly more threatening and challenging than the rest condition (all p’s < 0.005). Further, the PA index was significantly higher in the TSST than in all other conditions (all p’s < 0.001), indicating that the TSST was evaluated as the greatest threat or challenge among all conditions. The highest PA scores were found in the TSST, followed by Ergometer,

232

N. Skoluda et al.

(A) Cortisol in nmol/ml

20 stress / rest

VAS stress

30

10

12 10 8 6 4 2 0

0

n) n) in) in) i n) in ) in) in) mi mi 1 m 15 m 30 m 45 m 60 m 75 m -3 + ( -2 0 ( + + + + + ( ( ( ( s I( e ss II ( III IV V lin -stre stres ery very ery very se ery e ov o s tov ba o ov pr c c c c e rec po e r r re re

n) n) in ) in) mi mi 5m 0m -3 (+1 +1 +3 (-2 ( ( s( s I I s e I es li n tre er y er y str se e-s ov ov stba pr r ec po r ec i 0m

n)

(C)

140

(D)

160

120

80 60 40 20

HR in bpm

140 100 stress / rest

Alpha-amylase in U/ml

(B)

14

stress / rest

40

120 100 80 60 pre-stress

0 n) n) in) in) i n) in ) in ) in ) mi mi 1 m 1 5 m 3 0 m 45 m 6 0 m 7 5 m -3 + ( -20 ( + + + + + ( ( ( ( ( ( s I ss e II I II V IV es lin -stre ery very str se ery very e ry e ov o stov ba ov o c pr c c c e re c po e r r re re

stress / rest

post-stress

0 1

2

3

4

5

6

7

8

9

10

11

rest Stroop CPT TSST Ergometer

Figure 2 Variation over time in the respective conditions: (A) VAS stress, (B) cortisol, (C) alpha-amylase (sAA), and (D) heart rate (HR). The ‘‘stress/rest’’ bar indicates the stress/rest period (3 min during the CPT, 10 min during all other conditions). Descriptions: rest = no-stress control condition, CPT = cold pressor test, TSST = Trier Social Stress Test, Ergometer = bicycle ergometer task.

Stroop, and CPT. For SA, there was a significant effect of condition (F(4, 72) = 6.578, p < 0.001, Á2 = 0.268). Post-hoc test revealed significantly lower SA scores in the CPT compared to all other conditions, including the rest condition (all p’s < 0.005), suggesting that subjects perceived themselves as less capable to deal with the CPT and felt less control over the CPT. The highest SA scores were found in the Stroop, followed by TSST, Ergometer, rest condition, and CPT. For the PASA stress index, the repeated measures ANOVA revealed significant differences among conditions (F(4, 72) = 22.375, p < 0.001, Á2 = 0.554). Post-hoc tests showed that the TSST was anticipated as being significantly more stressful than all other conditions (all p’s < 0.001), followed by Ergometer, CPT, and Stroop. Similar to the findings reported above, participants reported different perceived stress levels among conditions (F(2.53, 45.51) = 9.523, p < 0.001, Á2 = 0.346), with each stressor inducing significantly higher perceived stress than the rest condition (all p’s < 0.001; see Figure 3A). The highest VAS delta values were found in the TSST, followed by Ergometer, Stroop, and CPT.

Participants showed different cortisol responses among conditions (F(4, 72) = 15.707, p < 0.001, Á2 = 0.466), with significantly higher cortisol increases in TSST and Ergometer than in the rest conditions (p < 0.001 and p < 0.005, respectively). In addition, higher cortisol increases were found in the TSST than in other conditions (all p’s < 0.005). In the ranking, the TSST was followed by Ergometer and CPT, while the Stroop failed to induce a significant cortisol increase (see Figure 3B). We computed responder rates for each stressor separately, according to which cortisol increases of 1.5 nmol/l and higher are classified as a significant response (Miller et al., 2013). Considering delta scores based on expected peaks, highest responder rates were found in the TSST (78.9%), followed by Ergometer (15.8%), Stroop (15.8%), and CPT (10.5%). The relative responder rates even increases when considering individual peaks, with highest responder rates in the TSST (94.7%), followed by Ergometer (47.7%), CPT (31.6%), and Stroop (26.3%). SAA responses differed significantly among the conditions (F(4, 72) = 7.208, p < 0.001, Á2 = 0.286). A post-hoc test revealed that sAA activity was only higher than in the rest

233

25 a**

20 15

VAS delta

(A) a**

a**

10

a**

5 0 -5

Cortisol delta increase in nmol/l

Stressor intensity & psycho-physiological responses

(B)

a** 8 6 a* b*

4 b** 2 b**

b**

rest

Stroop

0 -2

-10 Stroop

CPT

TSST

Ergometer

(C)

100 a** 80

a** c*

60 40 20 0

CPT

TSST

Ergometer

a** 100

HR delta increase in bpm

rest

Alpha-amylase delta increase in U/ml

10

(D)

80 60 40 a** c**

a** c** 20

c**

c**

0 rest

Stroop

CPT

TSST

Ergometer

rest

Stroop

CPT

TSST

Ergometer

Figure 3 Delta increase. Relative changes in the respective condition are indicated by delta increase values: (A) VAS stress (difference score post-stress and baseline), (B) cortisol (difference score recovery I and baseline), (C) alpha-amylase (sAA; difference score post-stress and baseline), and (D) heart rate (HR; difference score stress and pre-stress); Note: a = comparison with rest, b = comparison with TSST, c = comparison with Ergometer, * < 0.005, ** < 0.001 pairwise t-test with Bonferroni correction (overall, 10 pairwise comparisons). Descriptions: rest = no-stress control condition, CPT = cold pressor test, TSST = Trier Social Stress Test, Ergometer = bicycle ergometer task.

condition following Ergometer and TSST (both p’s < 0.001). Higher increases were found in the Ergometer than in the Stroop (p < 0.005). Using ln-transformed data, the highest sAA increases were found in Ergometer, followed by TSST, CPT, and Stroop; this ranking is in contrast to the ranking based on non-transformed data, with a position switch of CPT and Stroop (see Figure 3C). There was a significant effect of condition with regard to the changes in HR (F(2.31, 41.56) = 196.158, p < 0.001, Á2 = 0.916). All stressors, with the exception of the CPT, induced significantly higher heart rate increases compared to the rest condition (all p’s < 0.001). The increase in HR in response to the Ergometer was significantly higher than that in all other conditions (all p’s < 0.001), followed by TSST, Stroop, and CPT. All analyses described above were repeated using individual peaks (see supplementary material III). That is, delta scores were not computed based on the expected peak, but rather the actual peak within 30 min after the condition. Frequency analyses regarding the time point of the peak revealed that the majority of individual peaks corresponded to expected peaks. Repeated measures ANOVAs and post-hoc tests using individual peaks confirmed all findings with the exception of cortisol and sAA. In terms of individual peak, there was also a slight increase in cortisol levels in the

Stroop, which was not initially detected because most of the participants reached the highest cortisol levels immediately after the Stroop, and not 15 min later as expected. Using individual peaks, cortisol delta scores in Ergometer were no longer significantly different from either rest condition or TSST. Contrary to the findings based on expected peaks, sAA increases following both Ergometer and TSST were higher than those in the Stroop (but were no longer higher than in the rest condition).

4. Discussion This study was designed to investigate whether four commonly used laboratory stressors (Stroop, CPT, TSST, and Ergometer) differ in the likelihood and magnitude of psycho-physiological stress responses (‘stimulus response specificity’). We found significantly different time courses among the stress tests for all stress measures, thus providing evidence that the stressors indeed differ in their stress response patterns. There was a significant effect of condition, also confirmed by significantly different AUCG among conditions, in all stress measures except for sAA. With regard to psychological stress measures, we compared the conditions regarding anticipatory stress

234 appraisals immediately before stress exposure/rest, as well as pre-post-stressor changes in perceived stress (i.e., delta values). Compared to all other conditions, participants reported the highest anticipatory stress appraisals when exposed to the TSST. Each stressor in the current study induced an increase (i.e., delta values based on both expected and individual peaks) in perceived stress, which differed from the rest condition, with highest increases in the TSST, followed by Ergometer, Stroop, and CPT. Cortisol delta values increased in each stress condition, except for the Stroop when analyses were based on expected peak values. By far the highest cortisol increase compared to the rest condition was found in response to the TSST (which is underpinned by cortisol responder rates), followed by increases in Ergometer and CPT. With respect to autonomic markers, both sAA and HR delta values, based on expected peaks (using non-transformed data), increased in all stress conditions, with the highest increase in the Ergometer, followed by TSST, Stroop, and CPT. To note, when using ln-transformed data, descriptively lowest sAA increases were found in the Stroop, followed by the CPT. There might be a statistical explanation for this position switch in the ranking; the ln-transformation might reduce the heightened variance in the Stroop condition and thus resulted in a decreased mean. Whilst sAA increases in both Ergometer and TSST differed significantly from the rest condition, HR increases differed significantly from the rest condition in all stressors except for the CPT. These findings were also true for HR when considering individual peaks; however, sAA showed a slightly different picture. Considering individual peaks, significantly higher sAA increases in both Ergometer and TSST were found compared to the Stroop, but no longer compared to the rest condition. Since sAA is thought to be sensitive to and respond quickly to environmental changes, it cannot be completely ruled out that unknown environmental influences might have affected sAA levels. The finding that TSST exposure led to the highest increases in anticipatory stress appraisals indicates that a situation will only be perceived as stressful if it is appraised as threatening or challenging (primary appraisal) and if an individual’s resources to cope with the situation are evaluated as insufficient (secondary appraisal) (Lazarus and Folkman, 1984). As suggested by Dickerson and Kemeny (2004), the TSST combines those features (socialevaluative threat, uncontrollability) which inevitably lead to stress in most individuals. The other stressors may not be as threatening as the TSST, since these do not include the above-mentioned characteristics. Moreover, the egoinvolvement and thus the potential threat to the self which is experienced in the TSST are more salient than in the other stress protocols, since the speech task instruction requires participants to talk about their own personality (job application-related topics). With respect to perceived stress, it is interesting to note that even the Ergometer, a predominantly physiological stressor, was associated with an increase. For the Ergometer task, participants were instructed to cycle at a relatively high level, i.e., between 80% and 90% of self-rated effort, for 8 min. Experiencing exertion due to physical activity obviously resulted in feelings of ‘being stressed’. Importantly, this finding indicates a lack of stressor-specificity as well as heterogeneity in the meaning of the term ‘stress’

N. Skoluda et al. across stressors as used by participants. While this is of some utility when researchers wish to assess an unspecific feeling of ‘being stressed’, researchers should avoid using the term ‘stress’ if they are interested in stressor-specific subjective-emotional response profiles. Cortisol findings were in line with the findings of the meta-analysis by Dickerson and Kemeny (2004), with the highest responses to psychosocial stressors. There was a relatively small (but significant) cortisol response to the Ergometer compared to a psychosocial stressor. Previous studies have shown that intensity and duration of exercise influence the cortisol secretion (Davies and Few, 1973). It is conceivable that a more intense or exhaustive exercise protocol may have resulted in more pronounced cortisol increases. Other exercise protocols in which participants are asked to exercise until exhaustion may be more likely to elicit a significant HPA axis response, but such protocols suffer from a lack of standardization of exercise duration, since individuals differ in terms of the time when they report exhaustion. The finding that the highest autonomic stress responses were found in the Ergometer appears reasonable because physically-demanding exercises activate various vital systems, including cardiovascular and respiratory systems indicated by dose-response increases in heart rate and oxygen consumption (VO2 ) (Mann et al., 2013). In comparison to the HPA axis, the ANS is a fast-responding system and appears to be equally sensitive to both psychological and physiological stressors. This general ANS activation in a stressful situation leads to increased arousal and energy mobilization, which may facilitate coping with the current challenge, regardless of whether the stressor is psychological (e.g., the TSST) or physiological (e.g., the Ergometer) in nature (Chrousos and Gold, 1992). The overall findings suggest that tasks characteristics may be related to certain dimensions of the psychobiological stress response. That is, physiological stressors such the Ergometer may particularly activate the ANS, whereas psychosocial stressors such as the TSST may predominately stimulate the HPA axis. Frankenhaeuser and colleagues provided another possible explanation suggesting that the type of subjective stress response (feelings of distress vs. effort) may be associated with physiological activation. According to the authors, distress-causing tasks (such as the TSST) are related to HPA axis activation, while effort-demanding tasks (such as the Ergometer) are associated with ANS activation (Lundberg and Frankenhaeuser, 1980). We reviewed previous studies which used at least two stressors within the same study (listed in supplementary material I). The interpretation of these previous findings regarding comparability of stressors, however, is difficult due to methodological limitations. For example, it is difficult to compare stressors regarding stress responses when subjects are exposed to several stressors within the same session (multiple stressors) or when comparing stress responses of participants who are exposed to different stressors (between-subjects design). Furthermore, comparing only two stressors or failing to include a comparison with a non-stress condition may only allow for limited conclusions. In addition, the great majority of the reviewed studies assessed only one single stress measure. A multi-dimensional

Stressor intensity & psycho-physiological responses approach, as taken in the current study, enables researchers to investigate the interactions of psychological and physiological aspects of the stress system (Andrews et al., 2013) and to identify biological abnormalities (as indicated, for example, by the asymmetry of ANS and HPA axis reactivity under acute stress (see e.g., Gordis et al., 2008) or ‘sAA over cortisol ratio’ (Ali and Pruessner, 2012)). In this study, we used a multidimensional approach, which enables the assessment of psycho-physiological stress, i.e., measures of perceived stress, HPA axis (cortisol), and ANS (sAA, HR). This allows for the investigation of the stressors’ specific stress response pattern, i.e., whether and which stressors may result in stress responses. Our findings showed that laboratory stressors do indeed differ in their elicited stress response, depending on the stress measure of interest. That is, laboratory stressors may be comparable in elicited stress intensity in one measure but may differ from each other in another measure. We were only able to identify one out of 125 studies which used a multidimensional approach (i.e., using both multiple psychological and physiological stress measures) for the comparison of single laboratory stressors (see supplementary material I). In this study, subjective stress (VAS ‘distress’) and plasma cortisol were repeatedly obtained in participants who were exposed to both a CPT and a TSST on separate morning sessions (McRae et al., 2006). Stressors were equally associated with subjective stress, indicating that stressors did not differ in their stressor intensity regarding the elicited subjective stress which we also found in our study. Our study replicated the finding of higher cortisol responses to the TSST compared to the CPT. Summarizing the findings across the two studies, socialevaluative stressors appear to result in significantly higher cortisol stress responses compared to stressors without this characteristic. Several limitations need to be acknowledged. First, the CPT lasted for 3 min, whereas the other conditions lasted for 10 min. This duration of the CPT was chosen for ethical reasons because the immersion of a hand in ice-cold water for longer than 3 min may be harmful. However, we attempted to overcome this limitation by taking the different total duration of the stressors into account, thus holding relative proportions of the sample time points constant among conditions. Second, since exclusively young men were investigated, findings cannot be generalized to women or to individuals of older age. Future studies may investigate whether our findings are also observed in women, but these will need to address the influence of menstrual cycle phase. Future studies may also compare stress responses across different age groups to investigate the role of age. Third, we integrated an anticipation period for each stress protocol and rest condition, which was originally a component of the TSST only. While this constitutes a deviation from other stress protocols, it provided us with the opportunity to investigate anticipatory stress appraisals for each condition in a within-subjects design. Fourth, previous studies have shown that sAA activity might be influenced by the method of saliva collection, with stimulated saliva using Salivette possibly altering salivary flow rate and relative proportion where sAA is secreted (Rohleder and Nater, 2009). Fifth, although the multidimensional approach is a strength of the current study, one may argue that this may not go far enough. For example, there is ample evidence for the

235 stress-sensitivity of the immune system indicated by changes in inflammatory markers (for review see Steptoe et al., 2007). Also, there is some evidence from in vitro studies that immune cells are sensitive to glucocorticoids and catecholamines (e.g., Rohleder et al., 2010; Strahler et al., 2014). The assessment of glucocorticoid and catecholamine sensitivity may provide an approach to investigate how the ANS and HPA axis interact with the immune system which should be examined in future studies. Lastly, in this study, we disregarded ‘individual response specificity’, i.e., interindividual differences in stress responses. However, some of the analyses were repeated using individual peaks, allowing us to consider individual differences regarding time courses of psycho-physiological stress responses.

5. Conclusion Our findings show that various stress protocols, i.e., Stroop, CPT, TSST, and Ergometer, are able to induce psychological stress. However, stressors differed in the likelihood and magnitude of eliciting physiological stress responses. In particular, stress protocols which are predominantly physically demanding (such as exercise) are more likely to evoke autonomic stress responses, whereas uncontrollable and social-evaluative threatening stress protocols (TSST) are more suitable for eliciting both HPA axis and autonomic stress responses. The decision about which stress protocol is appropriate when planning a study depends on (a) the research questions, i.e., what dependent variable is of major interest, (b) the type of sample, and (c) the available resources (equipment and manpower). A multitude of stress protocols can be used to induce psychological stress. However, stress protocols need to meet certain requirements to also induce physiological responses. Our findings underpin the notion that stress protocols with social-evaluative, subjectively uncontrollable as well as physically demanding components are able to provoke increases in both HPA axis and autonomic activity. We showed that a stressor such as the CPT may lead to psychological responses, but fail to induce either HPA axis (salivary cortisol) or autonomic (sAA, HR) stress responses. However, a social-evaluative version of the CPT (SECPT, Schwabe et al., 2008) as well as the ‘Maastricht Acute Stress Test’ (MAST, Smeets et al., 2012), which both combine physical stress and a social-evaluative component, have proven to be effective in inducing cortisol increases. However, the sample in which potential stress responses are studied is of critical importance: For example, for health reasons, individuals with serious heart conditions should not be exposed to intense physical exertion such as Ergometer. Furthermore, available resources should be taken into account. Some stressors require special apparatus (e.g. stationary bicycle) or manpower (TSST audience usually consisting of two research assistants), or training of the research assistants (training for TSST audience). In sum, this study has shown that the choice of a stress protocol may determine the likelihood and magnitude of the induced psycho-physiological stress response (‘stimulus response specificity’). Our findings will hopefully guide future research in terms of making an adequate and

236 informed decision about choosing a suitable stress protocol, depending on the research question.

Role of funding sources The funding sources had no role in the design of the study, data collection and analysis, or drafting of the manuscript.

Conflict of interest The authors declare no financial interest related to the study.

Acknowledgments This research was partly funded by a grant of the Helene Bieber-Funds (fund sponsored by University of Zurich, Switzerland) (to UMN). UMN, JS, and NS gratefully acknowledge the support of the Volkswagen Foundation (Germany; AZ.:II/84 905).

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10. 1016/j.psyneuen.2014.10.002.

References Ali, N., Pruessner, J.C., 2012. The salivary alpha amylase over cortisol ratio as a marker to assess dysregulations of the stress systems. Physiol. Behav. 106 (1), 65—72. Allgrove, J.E., Gomes, E., Hough, J., Gleeson, M., 2008. Effects of exercise intensity on salivary antimicrobial proteins and markers of stress in active men. J. Sports Sci. 26 (6), 653—661. Andrews, J., Ali, N., Pruessner, J.C., 2013. Reflections on the interaction of psychogenic stress systems in humans: the stress coherence/compensation model. Psychoneuroendocrinology 38 (7), 947—961. Biondi, M., Picardi, A., 1999. Psychological stress and neuroendocrine function in humans: the last two decades of research. Psychother. Psychosom. 68 (3), 114—150. Borg, E., Kaijser, L., 2006. A comparison between three rating scales for perceived exertion and two different work tests. Scand. J. Med. Sci. Sports 16 (1), 57—69. Bosch, J.A., de Geus, E.J., Carroll, D., Goedhart, A.D., Anane, L.A., van Zanten, J.J., et al., 2009. A general enhancement of autonomic and cortisol responses during social evaluative threat. Psychosom. Med. 71 (8), 877—885. Chrousos, G.P., 2009. Stress and disorders of the stress system. Nat. Rev. Endocrinol. 5 (7), 374—381. Chrousos, G.P., Gold, P.W., 1992. The concepts of stress and stress system disorders. Overview of physical and behavioral homeostasis. J. Am. Med Assoc. 267 (9), 1244—1252. Davies, C.T., Few, J.D., 1973. Effects of exercise on adrenocortical function. J. Appl. Physiol. 35 (6), 887—891. Dickerson, S.S., Kemeny, M.E., 2004. Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research. Psychol. Bull. 130 (3), 355—391. Gaab, J., Rohleder, N., Nater, U.M., Ehlert, U., 2005. Psychological determinants of the cortisol stress response: the role of anticipatory cognitive appraisal. Psychoneuroendocrinology 30 (6), 599—610.

N. Skoluda et al. Gerra, G., Zaimovic, A., Mascetti, G.G., Gardini, S., Zambelli, U., Timpano, M., et al., 2001. Neuroendocrine responses to experimentally-induced psychological stress in healthy humans. Psychoneuroendocrinology 26 (1), 91—107. Gordis, E.B., Granger, D.A., Susman, E.J., Trickett, P.K., 2008. Salivary alpha amylase-cortisol asymmetry in maltreated youth. Horm. Behav. 53 (1), 96—103. Hamer, M., Steptoe, A., 2012. Cortisol responses to mental stress and incident hypertension in healthy men and women. J. Clin. Endocrinol. Metab. 97 (1), E29—E34. Kelly, C.B., Cooper, S.J., 1998. Plasma norepinephrine response to a cold pressor test in subtypes of depressive illness. Psychiatry Res. 81 (1), 39—50. Kirschbaum, C., Pirke, K.M., Hellhammer, D.H., 1993. The ‘Trier Social Stress Test’—–a tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology 28 (1—2), 76—81. Lazarus, R.S., Folkman, S., 1984. Stress, Appraisal, and Coping. Springer, New York. Lundberg, U., Frankenhaeuser, M., 1980. Pituitary-adrenal and sympathetic-adrenal correlates of distress and effort. J. Psychosom. Res. 24, 125—130. Mann, T., Lamberts, R.P., Lambert, M.I., 2013. Methods of prescribing relative exercise intensity: physiological and practical considerations. Sports Med. 43 (7), 613—625 (Research Support, Non-U.S. Gov’t Review). McRae, A.L., Saladin, M.E., Brady, K.T., Upadhyaya, H., Back, S.E., Timmerman, M.A., 2006. Stress reactivity: biological and subjective responses to the cold pressor and trier social stressors. Hum. Psychopharmacol. 21 (6), 377—385. Miller, R., Plessow, F., Kirschbaum, C., Stalder, T., 2013. Classification criteria for distinguishing cortisol responders from nonresponders to psychosocial stress: evaluation of salivary cortisol pulse detection in panel designs. Psychosom. Med. 75 (9), 832—840. Pruessner, J.C., Kirschbaum, C., Meinlschmid, G., Hellhammer, D.H., 2003. Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent change. Psychoneuroendocrinology 28 (7), 916—931. Rohleder, N., Nater, U.M., 2009. Determinants of salivary alpha-amylase in humans and methodological considerations. Psychoneuroendocrinology 34 (4), 469—485. Rohleder, N., Wolf, J.M., Wolf, O.T., 2010. Glucocorticoid sensitivity of cognitive and inflammatory processes in depression and posttraumatic stress disorder. Neurosci. Biobehav. Rev. 35 (1), 104—114 (Review). Schlotz, W., 2013. Stress reactivity. In: Gellman, M.D., Turner, J.D. (Eds.), Encyclopedia of Behavioral Medicine. Springer, New York, pp. 1891—1894. Schwabe, L., Haddad, L., Schachinger, H., 2008. HPA axis activation by a socially evaluated cold-pressor test. Psychoneuroendocrinology 33 (6), 890—895. Smeets, T., Cornelisse, S., Quaedflieg, C.W., Meyer, T., Jelicic, M., Merckelbach, H., 2012. Introducing the Maastricht Acute Stress Test (MAST): a quick and non-invasive approach to elicit robust autonomic and glucocorticoid stress responses. Psychoneuroendocrinology 37 (12), 1998—2008. Steptoe, A., Hamer, M., Chida, Y., 2007. The effects of acute psychological stress on circulating inflammatory factors in humans: a review and meta-analysis. Brain Behav. Immun. 21 (7), 901—912. Strahler, J., Rohleder, N., Wolf, J.M., 2014. Acute psychosocial stress induces differential short-term changes in catecholamine sensitivity of stimulated inflammatory cytokine production. Brain Behav. Immun., http://dx.doi.org/10.1016/j.bbi. 2014.07.014 (in press). Stroop, J.R., 1935. Studies of interference in serial verbal reactions. J. Exp. Psychol. 18, 643—662.